4.2. Compressor surge behavior
It has been shown that the compressor surging under standard maneuvering conditions is not critical. No compressor surge could be observed when load decreases from 100%-load to 70%-load as well as from 100%-load to 40%-load. Nevertheless, the last case could result in surge if the nominal operation line is too close to the surge line of the compressor map. In the case of engine shutdown (100%-0%-load) compressor surge will most probably occur [6][8]. The simulations also indicated that a fast increase in the fuel index may lead to surge.
In effect, the following steps have to be followed in order to avoid compressor surge in heavy weather conditions as well as during engine acceleration and deceleration phases:
・ Nominal operation line reasonably far from the surge line (proper turbocharger matching)
・ Turbocharger speed at nominal and overload operation to be within the turbocharger speed limits defined by the turbocharger manufacturer.
Conclusively, in the ACME Governor/ECU a maximum limit on the fuel index increase rate is applied to the output value of the PI speed regulator, in order to reduce the risk of surge.
4.3. Feedforward engine control
Incorporation of the acceleration signal in the TP algorithm is employed in order to implement an online assessment of the engine-ship interaction dynamics, that is required for the improvement of the feedforward engine control [17][18]. The feedforward control of a plant relies on the accurate modeling of the plant's "external disturbance", which in the case of a ship propulsion engine is the propeller torque fluctuation. Then the feedforward control action is generated, which in combination with the feedback (regulatory) plant control, comprises the overall plant control, by use of which certain performance specifications are met.
In the framework of the ACME project, the major objective in introducing the TP algorithm for advanced engine feedforward control, was to achieve maximum attenuation of disturbance (i.e. propeller load fluctuation) combined with smoother engine running, compared to the case in which only feedback control is used for disturbance attenuation. The above concept is illustrated in the following block diagram, Figure 9.
4.4. Sea-passage full-scale tests
In Figure 10 two typical prediction events from the 3 GB data pool collected during the ACME full-scale shipboard tests are presented. It can thus be seen that there is acceptable agreement between measured and predicted shaft torque profiles.
As demonstrated through full-scale testing, the accuracy of the existing TPA is moderate but can be used for improved control of main engine, preventing that the engine operating point is brought too close to the emergency shutdown value, under heavy weather conditions.
In the plots of Figure 11, typical signal fragments compiled out of the 3 GB TP System recordings are presented. The signals cover both propulsion and engine specific variables. The two (2) fragments shown have an approximate duration of 15 min each and are selected from the 5 hard disks (HDD units) that were used for logging onboard "Shanghai Express".
In the plots, shafting system variables and engine specific variables respectively are combined in order to give a sense of the sea-trials. Note that speed and torque, as well as fuel index, are zero for a time interval and then demonstrate a "spike". This corresponds most probably to engine reversing. However this is not indicated by negative speed values as the tacho sensor used was not sensitive to the sense of rotation. Therefore, a series of consequent engine reversing is presented as indicated by the spikes in speed, torque and index plots. The engine is finally stabilized at the speed of 30 rpm.
The major conclusions to be drawn from the analysis of the data collected including the above fragments are the following
1. The ACME Governor can successfully control the engine in steady-state, transient and manoeuving situations as fully demonstrated in the sea-trials.
2. The vertical aft ship acceleration signal is prone to noise originating from the hull structural and other vibrations. Extensive analogue and digital filtering of the signal was not successful in completely resolving this problem.
3 . Turbocharger surging is not a critical issue and this can be, at least partly, attributed to the incorporation of high speed limiters to the ACME Governor.
In order to resolve the difficulties in accurately measuring ship vertical acceleration, the development of an "acceleration-free" version of the TPA has been attempted by NTUA. The direction was to achieve modeling of the disturbance (i.e. propeller load torque fluctuation) based only on the information included in the propeller shaft torque signal itself. This is possible, because both the towing tank experiments and the full-scale tests have demonstrated that statistical analysis of the torque history can allow for increased predictability of the propeller torque demand in a time horizon of a few seconds or even tens of seconds continuously.
The proposed methodology for full-scale shipboard testing is demonstrated in Figure 12. Also, in Figures 13 and 14 two illustrative photos from the onboard installation works and modifications are given.
4.5. Final conclusion
The ACME project demonstrated that it is possible to apply advanced model-based, adaptive control strategies on marine Diesel propulsion plants in order to:
l) Extend capabilities of the plant, especially under adverse operating conditions, and
2) Improve performance, reliability and safety of the installation.
5. ACKNOWLEDGEMENTS
The present work was partly supported by the European Commission DGXII-Brite-Euram III programme, Project ACME (Adaptive Control of Marine Engines), contract BRPR-CT-95-0091, project number BR PR-CT95-0091, with Mr. E. Campogrande and, subsequently, Mr. F. Sgarbi as the CEC project officers. The partners of the ACME project were: DANAOS, NTUNLME, MAN-B&W, ABB, HAPAG LLOYD, HSVA and LIPS. The authors would like to thank their ACME partners and especially Messrs. D. Tsalapatis and K. Jensen of MAN-B&W.
7. REFERENCES
[1] J Benkelman W., Buitenhek M., "Full scale measurements and predicted sea keeping performance on the containership 'ATLANTIC CROWN "', Shipbuilding Laboratory, Netherlands Ship Research Centre TNO, Delft, the Netherlands, 1975
[2] Nakamura S,, Naito S, "Propulsive performance of a container ship in waves", J. S. N. A. Kansai Japan, 1976
